Efficient and reliable training of neural networks

Author

Yu, Hao ; Wilamowski, Bogdan M.

Author_Institution

Electr. & Comput. Eng., Auburn Univ., Auburn, AL

fYear

2009

fDate

21-23 May 2009

Firstpage

109

Lastpage

115

Abstract

This paper introduces a neural network training tool, NBN 2.0, which is developed based on neuron by neuron computing method. Error backpropagation (EBP) algorithm, Levenberg Marquardt (LM) algorithm and its improved versions are implemented in two different computing methods, traditional forward-backward computation and newly developed forward-only computation. The software can handle not only conventional multilayer perceptron (MLP) networks, but also arbitrarily connected neuron (ACN) networks. Several examples are presented to explain how to use this tool for neural network training. The software is developed based on Visual Studio platform using C++ language and it is available for everyone on the Web site.

Keywords

C++ language; learning (artificial intelligence); neural nets; visual programming; C++ language; Levenberg Marquardt algorithm; NBN 2.0; Visual Studio platform; arbitrarily connected neuron network; error backpropagation algorithm; neural network training; neuron computing method; Application software; Backpropagation algorithms; Computer networks; Industrial training; Jacobian matrices; Multilayer perceptrons; Network topology; Neural networks; Neurons; Software tools; neural networks; training tool;

fLanguage

English

Publisher

ieee

Conference_Titel

Human System Interactions, 2009. HSI '09. 2nd Conference on

Conference_Location

Catania

Print_ISBN

978-1-4244-3959-1

Electronic_ISBN

978-1-4244-3960-7

Type

conf

DOI

10.1109/HSI.2009.5090963

Filename

5090963